- Email: [email protected]
Estimating higher education induced energy consumption: The case of Northern Cyprus
Energy 66 (2014) 831e838
Contents lists available at ScienceDirect
Energy journal homepage: www.elsevier.com/locate/energy
Estimating higher education induced energy consumption: The case of Northern Cyprus lu* Salih Turan Katirciog Department of Banking and Finance, Eastern Mediterranean University, P.O. Box 95, Famagusta, Northern Cyprus, Via Mersin 10, Turkey
a r t i c l e i n f o
a b s t r a c t
Article history: Received 9 July 2013 Received in revised form 25 November 2013 Accepted 17 December 2013 Available online 22 January 2014
This study estimates higher education-induced energy consumption in the case of the TRNC (Turkish Republic of Northern Cyprus). Although the TRNC is a non-recognized state and a small island; it attracts many international students each year and has shown tremendous development in higher education since the 1990s. The results of the present study reveal that higher education development has an ongoing relationship with electricity consumption; electricity consumption reacts to its long-term equilibrium level by 80.95% as a result of higher education development. Finally, the results of the present study reveal that higher education development in Northern Cyprus exerts a positive and significant growth impact not only on electricity consumption but also on overall oil consumption, both in the short- and long-term; therefore, it can be inferred that higher education development in this small and non-recognized island state is a catalyst for the growth of energy consumption. Ó 2013 Elsevier Ltd. All rights reserved.
JEL classification: C22 C51 F20 I21 I23 Keywords: Higher education Energy consumption Bounds test Northern Cyprus
1. Introduction In addition to investigating the relationship between international trade expansion and economic growth, which has become a popular topic in development economics, the relationship between international tourism and economic growth has started to attract attention. Studies on the topic have been conducted by Katircioglu [1e4], Cortés-Jiménez and Pulina [5], Gunduz and Hatemi-J [6], Oh [7], Ongan and Demiroz [8], Dristakis [9], and Balaguer and Cantavella-Jordá [10]. International trade (including services and tourism) expansion can contribute to economic growth through various channels [11]. However, to the best of the author’s knowledge, the contribution of international tourism to economic growth and particular segments of the economy has received little interest from researchers to date. Therefore, an investigation of these channels in the case of international tourism is warranted. Higher education is an important global phenomenon. In fact, millions of people pursue their higher education at overseas
* Tel.: þ90 392 630 2008; fax: þ90 392 630 2032. E-mail address: [email protected]. 0360-5442/$ e see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.energy.2013.12.040
institutions each year. Thus, higher education can be considered as a type of student tourism that contributes to national income, employment, and the wealth of local citizens [1]. Davies & Lea [12] and Carr [13] suggest that university students are actually a distinct population, differing in age and socio-cultural, educational, and economic characteristics despite the fact that the majority of them might be chronologically defined as belonging to the youth population. Furthermore, Smeaton et al. [14] show that the university environment encourages students to travel and take holidays. The contribution of student tourism to the economy is of particular importance to developing countries. Stevens and Weale [15] mention that living standards in most countries, and especially those in Europe, have risen over the last millennium due to developments in education. It is obvious that one of the most important factors affecting the demand for private secondary or higher education is the household income level and the costs incurred by a family when it takes the decision to invest in education [16]. However, there are generally accepted social and economic factors that affect a household’s demand for education such as the parents’ education, the geographical location of the place of residence, the size and composition of the family, the occupation of the primary earner, and the family’s own consideration of its social
832
S.T. Katircioglu / Energy 66 (2014) 831e838
status [16]. On the other hand, there are some external factors that might also affect the decision to study abroad such as the political and economic conditions of the targeted country or region, geographical location of the targeted institution, student fees, scholarship opportunities, medium of instruction, and the accreditation of the diploma that is received from these institutions. From an economic viewpoint, an important reason for education is its impact on reducing inequalities of income [17] and the relationship between education and the labor market [16]. Some studies have focused on the estimation of the rate of return from education [18e20]. On the other hand, there are few studies in the field of higher education. Katircioglu [1] concludes that higher education acts as a catalyst for real income growth in a nonrecognized state, namely, the TRNC (the Turkish Republic of Northern Cyprus) in the Mediterranean. As previously mentioned, since millions of students pursue their higher education at overseas institutions, student tourism can be considered as a part of traditional (international) tourism [1]. The contribution of higher education development is not only important for the overall income of a nation but also for particular segments of the economy. The energy sector is an example of one of the various channels through which higher education contributes to expansion in the economy. A growth in international (student) tourism results in an increased demand for energy, for example through accommodation and transportation [21e23]. If higher education institutions progress considerably and are successful in attracting overseas students as well as local citizens, this development in higher education institutions will lead to increase in the number of buildings utilized, and therefore in energy demand. Increases in the number of higher education institutions and the number of overseas students will also lead to an increase in the number of businesses in the economy including restaurants, dormitories, travel agencies, and dry cleaners, among others. All of these will raise the energy (including electricity) demand in the economy; therefore, while higher education development contributes to the overall economy through different sectors, it also leads to a long-term expansion in energy consumption. This highlights the importance of adapting successful energy management policies in order to achieve energy conservation and environmental protection. Increased energy consumption will result in increases in energy expenditures by households and firms, which in turn will lead to an increase in aggregate income in the economy. In this respect, research on the nature of the relationship between higher education development and energy consumption deserves attention. Many studies in the existing literature focus on the relationship between real income growth and particular segments of the energy sector, including an early study by Kraft and Kraft [24]; and among the latest studies, Ouédraogo [25],1 Wolde-Rafael [26], Odhiambo [27], Apergis and Payne [28], and Lee [29]. However, little research has studied the interaction between the energy sector and particular segments of the economy. For example, there are few studies on energy consumption or the patterns of energy use in the case of international tourism. Of those in existence are Tabatchnaia [30], Gössling [31], Ceron and Dubois [32], Becken and Simmons [33], Becken et al. [22], Trung and Kumar [34], Warnken et al. [35], Becken et al. [21], and Nepal [36]. Recognizing the importance of this issue, the present study estimates higher education induced electricity consumption and investigates the empirical relationship between higher education and energy consumption, which is proxied by electricity and oil consumptions, in the TRNC. The TRNC is not recognized by countries other than its mainland, Turkey, but attracts many international
1
Ouédraogo [25] also presents a brief review of the literature in this field.
students from around the world as a result of the successful development of its higher education sector. A study by Katircioglu [1] validates the higher education-led Growth Hypothesis in the case of Northern Cyprus. The TRNC, which is located in a strategic location in the Eastern Mediterranean, was established in 1983 in an already divided island and has a population of over 286,973 and a per capita income of 14,421.77 US$ [37]. The TRNC does not have any foreign trade relationships with countries other than Turkey due to its political non-recognition. Therefore, international tourism and the emergence of the higher education sector are two major sources of foreign exchange for this small island state. However, the tourism sector also faces great difficulties in attracting international tourists because of problems such as the lack of direct flights to Northern Cyprus and high transportation costs. To the best of the author’s knowledge, there are currently no empirical studies investigating the relationship between higher education and energy consumption in the literature. Thus, this study is the first of its kind as it investigates the ongoing equilibrium relationship and direction of causality between higher education development and energy consumption in the TRNC. The services sector in Northern Cyprus was given priority as a result of the political isolation and embargoes faced by the country in every field. The 1980s became a transition period from the manufacturing industry to services with a focus on tourism and higher education. The tourism sector suffered also from political problems, so the island could not attract the amount of tourists needed to stimulate significant growth in the economy. Tourists from abroad were targeted by allowing the opening of casinos on the island. Now, many casinos have opened in Northern Cyprus and attract tourists from Turkey and the south of Cyprus. Legalized gambling is prohibited in both of these countries. In 2008, 808,682 tourists visited Northern Cyprus, of which 80% were from Turkey. Net tourism revenues constituted 10.88% of GDP (gross domestic product) in 2010 [37]. On the other hand, the demand for higher education in Northern Cyprus has increased considerably since the 1990s, mainly due to the number of students from Turkey and overseas advertising especially in Africa and the Middle East. There are eight universities in Northern Cyprus: The EMU (Eastern Mediterranean University ), which is the oldest and the largest, was established in 1979, NEU (Near East University), LEU (Lefke European University), GAU (Girne American University), CIA (Cyprus International University), the Northern Cyprus campus of METU (which is a Turkish university) (Middle East Technical University), the northern campus of ITU (which is a Turkish university) (Istanbul Technical University), and the UMK (University of Mediterranean Karpasia). At the beginning of the 2011e2012 academic year, there were almost 53,000 students studying at these eight universities, of which 20.40% were Turkish Cypriots, 72.95% were from mainland Turkey, and 6.65% were from various overseas countries [37]. Overseas students have been coming to Northern Cyprus for higher education since 1982. Since then, there has been a steady increase in the number of overseas students who now come from more than 68 different countries. The presence of internationally recognized and accredited universities in Northern Cyprus contributes to the positive image of the country in the international arena. Furthermore, the expansion of infrastructure and facilities at the universities of Northern Cyprus continues at an unprecedented rate and may now be compared favorably with their international counterparts. Therefore, the higher education sector is now the most important sector in Northern Cyprus and earns considerable foreign exchange. The sector is also contributing to the growth of this small and nonrecognized island state. The development of higher education in Northern Cyprus has led to considerable increase in energy demand in the island as well. The overall net electricity consumption in
S.T. Katircioglu / Energy 66 (2014) 831e838
Northern Cyprus is 887.7 million kWh, and since 2010, the net electricity consumption per capita has been 3093 kWh. On the other hand, the overall oil consumption in Northern Cyprus was 254,901 million tons in 2010. Of the overall electricity consumption in Northern Cyprus, almost 3.4% was used by higher education institutions in 2010.2 Electricity power is mainly distributed and priced in the forms of housing, trade, tourism, water motors, roads/streets, state offices, and the prosecutor’s office. Electricity in Northern Cyprus is distributed by two major firms, of which one is a private firm originating from Turkey. Petroleum products are also distributed by one public and one private firm on the island. Electricity in Northern Cyprus is produced by two main power stations where petroleum products are completely imported from abroad. In 2010, 20% of the overall imports in Northern Cyprus were petroleum products [37]. Energy consumption in the universities of North Cyprus is mainly categorized into electricity for buildings and fuel oil for heaters and transportation. The electricity consumption for buildings constitutes almost 60% of the overall higher education energy consumption while 30% is accounted for by fuel oil for heaters and 10% for fuel oil for transportation (cars and buses). This indicates that the main drivers of higher education energy consumption in Northern Cyprus are electricity and fuel oil. The present paper proceeds as follows: Section 2 presents the theoretical setting of the empirical methodology and Section 3 defines the data and the methodology of the study. Section 4 presents the results and discussion, and the paper concludes with Section 5.
2. Theoretical setting There are two thermal power plants in Northern Cyprus, both of which use mainly oil to drive them. The oil is completely imported from abroad in this small island and exchange rates are the major determinant not only of the flow of international students to universities in Northern Cyprus but also of the imported oil. With this respect, a starting point of the theoretical setting in the present study is that higher education development, oil use, and real exchange rates are likely to be a determinant of electricity consumption. Oil use has been added to equation (1) in order to avoid omitting a variable problem; thus, the following functional relationship has been put forward in the present study:
Et ¼ f ðOILt ; HEt ; RERt Þ
(1)
where electricity consumption (E) is a function of oil consumption (OIL), higher education (HE), and real exchange rates (RER) in equation (1). The functional relationship in equation (1) can be expressed in logarithmic form to capture the growth impacts:
ln Et ¼ b0 þ b1 lnOILt þ b2 lnHEt þ b3 lnRERt þ εt
(2)
where at period t, ln E is the natural log of electricity consumption; ln OIL is the natural log of oil consumption; ln HE is the natural log of the higher education variable; ln RER is the natural log of real exchange rates; and ε is the error disturbance. The dependent variable in equation (2) may not immediately adjust to the long-term equilibrium level following a change in any of the determinants. Therefore, the speed of adjustment between the short- and the long-term levels of the dependent variable (E)
2 Estimated from the SPO (State Planning Organization) [37] statistics and universities’ reports.
833
can be captured by estimating the following error correction model (ECM):
Dln Et ¼ b0 þ
n X
b1 Dln Etj þ
i¼1
þ
n X
n X
b2 DlnOILtj þ
i¼0
n X
b3 DlnHEtj
i¼0
b4 DlnRERtj þ b5 εt1 þ ut
i¼0
(3) where D represents a change in the ln E, ln OIL, ln HE, and ln RER variables and εt-1 is the one period lagged ECT (error correction term), which is estimated from equation (2). ECT in equation (3) shows how fast the disequilibrium between the short- and longterm values of the dependent variable is eliminated in each period. The expected sign of ECT is negative [See Ref. [38]].
3. Data and methodology The data used in this paper are annual figures covering the period 1979e2010 and the variables of the study are electricity consumption per capita in kWh (which is the acronym of kilowatt hour) (E), oil consumption per capita in liter (OIL), and total number of students studying at higher education institutions in Northern Cyprus (HE). The choice of electricity consumption and oil consumption in this study has been made based on the fact that data for overall energy use as kt of oil equivalent is not available for Northern Cyprus; electricity and oil are two major parts of the overall energy consumed on the island. The overall oil per capita data in this study includes (1) petroleum, (2) gas oil, (3) diesel, (4) fuel oil, (5) jet fuel (A-1), (6) liquefied petroleum gas, and (7) lube oil. As the oil consumed in Northern Cyprus is completely imported from abroad, exchange rates are also a major determinant not only of the number of international students coming to and studying on the island (higher education sector) but also of electricity consumption as mentioned previously. Some studies in the relevant literature advise the addition of exchange rates as control variables when considering tourist or foreign student arrivals, since it is a superior proxy for international prices. Examples of studies advising this are Balaguer & Cantavella-Jordá [10] and Katircioglu [2]. REE (real exchange rates) have been added to all of the models in the present study as a regressor in this respect and the data were gathered from the State Planning Organization of Northern Cyprus [37]. All of the variables have been transformed into their natural logarithm forms to capture the growth impacts as also mentioned in the previous section. ZA’ (Zivot and Andrews) [39] test for unit root has been employed in the present study to investigate the order of the integration of the variables as there are breaks in the series; thus, time series properties of the variables of the study are likely to be affected by these breaks. In the literature, Perron [40] was first to demonstrate that breaks in the series may affect the power of the unit-root tests in such a way that null of non-stationarity is underrejected. However, Zivot and Andrews [39] argue that under an alternative hypothesis, the breakpoint should be treated as unknown, as it is by Perron [40] who biases his results in favor of the rejection of the unit-root hypothesis. Zivot and Andrews [39] utilize the same modeling framework as Perron [37] but with an unknown breakpoint instead of a known breakpoint as in Perron [40]. Therefore, the ZA [39] unit root tests with breaks will be employed in the present study to investigate the integration level of the variables under consideration. To investigate the long-term relationship in equation (1), the bounds test within ARDL (the autoregressive distributed lag)
S.T. Katircioglu / Energy 66 (2014) 831e838
834
modeling approach was adopted. This approach was developed by Pesaran et al. [41] and can be applied irrespective of the order of the integration of the variables (irrespective of whether regressors are purely I(0), purely I(1), or mutually co-integrated). The ARDL modeling approach involves estimating the following:
Dln Et ¼ a0 þ
n X
bi Dln Eti þ
i¼1
þ
n X
n X
ci DlnOILti þ
i¼0
n X
di DlnHEti
i¼0
ei DlnRERti þ s1 ln Eti þ s2 lnOILti
i¼0
þ s3 lnHEti þ s4 lnRERti þ εt (4) In equation (4), D is the difference operator and εt is a serially independent random error with a mean of zero and a finite covariance matrix. The F-test is used for investigating a (single) long-term relationship in equation (4). In the case of a long-term relationship, the F-test indicates which variable should be normalized. In equation (4), when ln E is the dependent variable, the null hypothesis of no level relationship is H0: s1 ¼ s2 ¼ s3 ¼ s4 ¼ 0 and the alternative hypothesis of a level relationship is H1: s1 s s2 s s3 s s4 s 0. In the present study, the conditional ECM using the ARDL approach will be employed in the case of a level relationship to estimate the short-term coefficients plus the error correction term in addition to the long-term coefficients in equation (2). In addition, as also suggested by Pesaran et al. [41], the time series properties of the key variables (E, HE, and F) in the conditional ECM of the present study can be approximated by double-log error correction at lag levels that might be different for each explanatory variable, augmented with appropriate deterministics such as the intercept and time trend. Equation (3) of the present study will be used to estimate the conditional error correction model where b5 is the coefficient of the error correction term and is expected to be negative [38]. In the case of level relationships based on the bounds test, conditional Granger causality tests can be carried out using the error in the correction model. In doing so, the short-term deviations of the series from their long-term equilibrium path are again captured by including an error correction term. Therefore, similar to equation (3), conditional error correction models (that use the ARDL approach) for Granger causality can be specified as follows:
2
3
2
3
2
32
3
d11;1 d12;1 d13;1 d14;1 m1 Dln Et Dln Et1 6 DlnOILt 7 6 m2 7 6 d21;1 d22;1 d23;1 d24;1 76 DlnOILt1 7 6 7 6 7 6 76 7 4 DlnHEt 5 ¼ 4 m3 5 þ 4 d31;1 d32;1 d33;1 d34;1 54 DlnHEt1 5 d41;1 d42;1 d43;1 d44;1 DlnRERt1 m4 DlnRERt 2 32 3 d11;i d12;i d13;i d14;i Dln Eti 6 d21;i d22;i d23;i d24;i 76 DlnOILti 7 76 7 þ.þ6 4 d31;i d32;i d33;i d34;i 54 DlnHEti 5 d41;i d42;i d43;i d44;i DlnRERti 3 3 2 ε1;t 41 6 42 7 6 ε2;t 7 7 7 6 þ6 4 43 5ECTt1 þ 4 ε3;t 5 ε4;t 44 2
(5) In equation (5), D denotes the difference operator. ECTt-1 is the lagged error correction term derived from the long-term equilibrium model. Furthermore, ε1,t, ε2,t, ε3,t, ε4,t, and ε5,t are serially independent random errors with a mean of zero and a finite covariance matrix. Finally, according to the conditional ECM for
causality tests, having statistically significant t ratios for ECTt-1 in equation (5) would meet the conditions of long-term causations while significant F ratios for the overall models in the same equation would denote short-term causations. In the final step, impulse responses plus the variance decomposition for variables have been estimated. Impulse responses would denote how the series react to exogenous shocks in another over the periods while variance decompositions would determine how much of the forecast error variance of the dependent variable can be explained by exogenous shocks to independent variables. Impulse responses and variance decompositions are expected to be complementary to the ECM and causality tests. 4. Results and discussion Fig. 1 presents line graphs of the variables under consideration. It can be seen that various breaks are available in the series including for the ln HE variable, which was due to the latter effects of military intervention in Turkey during September 1980. Therefore, ZA [39] unit root tests would be suitable for testing the integration level of the series. Table 1 presents the ZA [39] unit root test results for the variables where various lag structures and selected breaking years are available. The null hypothesis of a unit root cannot be rejected in the cases of ln E and ln OIL at their level forms; but it can be rejected at the first differences. On the other hand, the null hypothesis of a unit root can be rejected in the cases of ln HE and ln RER at their level forms. Therefore, it is concluded that electricity consumption (ln E) and oil consumption (ln OIL) are integrated at order one, I(1), while the higher education variable (ln HE) and real exchange rates (ln RER) are integrated at order zero, I(0). Unit root tests have provided mixed results for the variables in this study; therefore, bounds tests will be employed to investigate the long-term equilibrium relationship in equation (1) using the ARDL modeling approach as suggested by Pesaran et al. [41]. In order to proceed with the bounds tests, the dependent variable needs to be integrated at order one3; therefore, this condition has already been satisfied. Critical values for F statistics for small samples are presented in Table 2, as taken from Narayan [42]. Table 3 gives the results of the bounds test for a level relationship in equation (1). These models are utilized under three different scenarios as suggested by Pesaran et al. [[41]:295e296], which are with restricted deterministic trends (FIV), with unrestricted deterministic trends (FV), and without deterministic trends (FIII). All of the intercepts in these scenarios are unrestricted.4 The results presented in Table 3 imply that the application of the bounds F-test using the ARDL modeling approach suggests a level relationship in equation (2) of the present study, as can be seen in Table 3. Optimum lag by the Schwartz criterion is one and the null hypothesis of H0: s1 ¼ s2 ¼ s3 ¼ s4 ¼ 0 in equation (4) can be rejected according to the bounds test results (as confirmed by FIII and FIV) (shown in Table 3), which are highly significant. Results from the FV scenario when an unrestricted trend and unrestricted intercept are included do not, however, enable us to reject H0: s1 ¼ s2 ¼ s3 ¼ s4 ¼ 0. The FIII and FIV scenarios also produced better results in the original study by Pesaran et al. [41]. Therefore, a level relationship exists in equation (1); that electricity consumption is in a long-term relationship with oil consumption, higher education, and real exchange rates. The results from the application of the bounds t-test in each ARDL model do allow for the imposition of the trend restrictions in the model since t-ratios are significant in the
3 4
Please refer to Pesaran et al. [41]. For detailed information, please refer to Pesaran et al. [41], pp. 295e296.
S.T. Katircioglu / Energy 66 (2014) 831e838
835
Fig. 1. Plots of ln E, ln OIL, ln HE, and ln RER.
case of the FIII scenario [See Ref. [41]: 312]. This proves that results from the FIV scenario are robust. Having a level relationship in the bounds tests allows for the adoption of the ARDL approach to estimate the level coefficients, as also discussed in Pesaran and Shin [43] and formulated in equation (2) of the present study. The resulting estimates of level relationships under the ARDL specification in the case of the Electricity Consumption Model (where electricity consumption is a dependent variable) (lag structures: 3, 1, 3, 2) in equation (2) can be given by: Electricity consumption model in equation (2):
significant (0.0741). This implies that a one percent change in higher education volume will lead to a 0.0741% change in the electricity volume in the same direction and in the short-term period. The direction of causality will be investigated using the conditional Granger causality tests and under the ARDL mechanism in both the short- and long-term periods. F-Statistics for short-term causations and t-statistics of the ECTs for long-term causations are given in Table 5, as estimated from equation (5).
^t ln Et ¼ 1:063ðlnOilt Þ 0:047ðlnHEt Þ 0:029ðlnRERt Þ þ1:726 þu
Table 1 ZA (1992) tests for unit root under a single structural break.
ð0:000Þ
ð0:278Þ
ð0:629Þ
ð0:338Þ
Statistics (level)
where ût is an error correction term and p-values are given in the parentheses. The estimated parameter of oil consumption (ln OIL) is statistically significant at the 0.01 level and elastic; oil use has an elastic, statistically significant, and positive impact on electricity consumption (1.063) in North Cyprus. The long-term coefficients of the higher education variable and real exchange rates are not statistically significant. In the next stage, the conditional ECM, which is associated with the above level relationship, should be estimated. This is presented in Table 4. The ECT term where electricity consumption is a dependent variable is high (0.8095), statistically significant, and negative.5 This implies that electricity consumption converges to its long-term equilibrium path by 80.95% speed of adjustment through the channels of higher education development, oil use, and real exchange rates. It is important to note that the short-term coefficient of the higher education variable is positive and statistically
5
ECT terms should be negative by expectation.
ln E Break year Lag length ln OIL Break year Lag length ln HE Break year Lag length ln RER Break year Lag length
Statistics (first difference)
ZAB
ZAT
ZAI
ZAB
ZAT
ZAI
2.815 2005 0 3.940 1992 1 4.634 2006 3 4.615 2003 0
2.368 1993 0 3.886 1993 1 4.760a 2006 3 3.951 2000 0
2.026 1990 0 3.643 1987 1 5.290a 1997 3 5.233a 2003 0
6.804 2004 0 5.124a 2005 4 4.282 1998 4 6.711 1991 0
5.534 2004 0 5.064 1988 3 4.389b 1999 4 6.192 1986 0
5.532 1998 0 4.841b 1987 3 3.656 1988 4 6.241 1985 0
Conclusion
I(1)
I(1)
I(0)
I(1)
E represents the electricity consumption per capita in KWS; OIL is overall oil consumption per capita in liters; HE represents the total number of university students in the higher education institutions of North Cyprus; RER stands for real exchange rates. All of the series are at their natural logarithms. ZAB represents the model with a break in both the trend and intercept; ZAT is the model with a break in the trend; ZAI is the model with a break in the intercept. Tests for unit roots have been carried out in E-VIEWS 7.2. a Rejection of the null hypothesis at the 5%. b Rejection of the null hypothesis at the 10%.
S.T. Katircioglu / Energy 66 (2014) 831e838
836 Table 2 Critical values for the ARDL modeling approach. k¼3
FIV FV FIII tV tIII
0.10
Table 4 The ARDL error correction model for energy consumption and higher education.
0.05
0.01
I(0)
I(1)
I(0)
I(1)
I(0)
I(1)
3.378 3.868 3.008 3.130 2.570
4.274 4.965 4.150 3.840 3.460
4.048 4.683 3.710 3.410 2.860
5.090 5.980 5.018 4.160 3.780
5.666 6.643 5.333 3.960 3.430
6.988 8.313 7.063 4.730 4.370
(1) k is the number of regressors for the dependent variable in the ARDL models, FIV represents the F statistic of the model with an unrestricted intercept and restricted trend, FV represents the F statistic of the model with an unrestricted intercept and trend, and FIII represents the F statistic of the model with an unrestricted intercept and no trend. (2) tV and tIII are the t ratios for testing s1 ¼ 0 in equation (4) respectively with and without a deterministic linear trend. Source: Narayan [42] for F-statistics and Pesaran et al. [41] for t-ratios.
The results in Table 5 reveal the unidirectional causality in the long-term period that runs from higher education development, oil use, and real exchange rates to a growth in electricity consumption; this is because the t-statistic of the ECT where electricity consumption is a dependent variable is statistically significant and negative (2.298). The other t ratios are not statistically significant for long-term estimations. It is concluded that higher education development and oil consumption are catalysts for the growth of energy consumption in the long-term period. In the case of oil consumption, on the other hand, there are statistically significant F statistics that denote some short-term causations. These statistics arise from real exchange rates in electricity consumption, electricity, higher education, and real exchange rates for oil consumption; this shows that higher education, electricity consumption, and real exchange rates are catalysts in the short-term period (as expected). It is again concluded that electricity consumption, higher education development, and exchange rates are also catalysts for the overall oil consumption in the short-term. Fig. 2 depicts line plots of impulse responses among the series. As can be seen from the figure, the response of electricity consumption to a shock in higher education is in the same direction (positive) over time, which, for example, suggests that an increase in higher education uptake in Northern Cyprus will also lead to an increase in electricity consumption. Furthermore, the effect of a shock in higher education on electricity consumption reaches its peak level in periods (years) 5 and 6. On the other hand, the Table 3 The bounds test for level relationships. Variables
With deterministic trends
Without deterministic trend
FIV
FIII
FV
tV
Conclusion
tIII H0
FE (ln E/ln OIL, ln HE, ln RER) p ¼ 1* 8.483c 2 8.527c 3 7.143c 4 4.079b
3.602a 2.028a 1.941a 2.194a
2.535a 13.255c 1.828a 13.348c 1.787a 11.296c 2.001a 6.376c
3.908c Rejected 3.713c 3.310b 3.517c
The Schwartz criterion (SC) has been used to select the optimum lag length required in the bounds test. p shows the lag levels and * denotes the optimum lag selection in the model as suggested by SC. FIV represents the F statistic of the model with an unrestricted intercept and restricted trend, FV represents the F statistic of the model with unrestricted intercept and trend, and FIII represents the F statistic of the model with an unrestricted intercept and no trend. tV and tIII are the t ratios for testing s1 ¼ 0 in equation (4) respectively with and without a deterministic linear trend. a Indicates that the statistic lies below the lower bound. b That it falls within the lower and upper bounds. c That it lies above the upper bound.
Dependent variable: electricity consumption (E) Lag structure: (3, 1, 3, 2) Regressor
Coefficient
Standard error
p-Value
ût-1 Dln Et-1 Dln Et-2 Dln OIL Dln OILt-1 Dln OILt-2 Dln HE Dln RER Dln RERt-1 Intercept
0.8095 0.2384 0.4746 0.4652 0.5169 0.3655 0.0741 0.0993 0.1186 0.0000
0.1263 0.1729 0.2019 0.0886 0.1177 0.1208 0.0358 0.0520 0.0596 0.0114
0.0000 0.1839 0.0297 0.0000 0.0003 0.0070 0.0524 0.0713 0.0616 1.0000
Adj. R2 ¼ 0.706, S.E. of Regr. ¼ 0.032. AIC ¼ 3.762, SBC ¼ 3.290. F-stat. ¼ 8.484, F-prob. ¼ 0.000. D-W stat. ¼ 2.060.
response of oil consumption to a shock in higher education is again positive and reaches its peak level in period (year) 4. Impulse response analyses suggest a positive and significant impact of higher education on both electricity and oil consumption. Finally, Table 6 shows the variance decomposition results among the series. The results show that in the initial periods, lower levels of the forecast error variance of electricity consumption are explained by exogenous shocks to higher education and the other variables. However, these ratios increase over time. For example, in period 10, 6.901% of the forecast error variance of electricity consumption is explained by exogenous shocks to higher education. The results in Table 6 show that forecast error variances of oil consumption, with respect to exogenous shocks in higher education, are higher compared to those in the cases of electricity consumption. This suggests that higher education exerts greater impacts on overall oil consumption than it does on electricity consumption. The results of the Granger causality tests, impulse responses, and variance decompositions reveal that development of higher education in Northern Cyprus exerts both significant and positive impacts on energy consumption (electricity and oil). 5. Conclusion This paper empirically estimates the higher education-induced energy consumption (namely electricity consumption per capita) in the case of Northern Cyprus, which is a small island and a nonrecognized state, but which attracts many international students each year from different regions of the world. The justification for conducting this research is the fact that energy consumption is likely to be affected by such a small island attracting so many international students. Furthermore, to the best of the author’s knowledge, this study is the first of its kind in the relevant literature. The results of the present study reveal that long-term equilibrium relationship exists between higher education development, oil use, and electricity consumption on this small island. Electricity consumption converges to its long-term equilibrium path by 80.95% speed of adjustment every year through the channels of higher education development and oil consumption. Finally, results from the conditional Granger causality tests, impulse response, and variance decomposition analyses suggest that higher education development in Northern Cyprus exerts a positive and significant growth impact not only on electricity consumption but also on overall oil consumption in both the short- and long-term; therefore, it can be inferred that the development of higher education in this small and non-recognized island state is a catalyst for an increase in energy consumption.
S.T. Katircioglu / Energy 66 (2014) 831e838
837
Table 5 Results of the conditional Granger causality tests. Dependent variable
F-statistics [probability values]
Dln Et Dln Et Dln HEt Dln OILt Dln RERt a b
e 2.169 [0.236] 7.272a [0.038] 2.000 [0.260]
Dln HEt 1.397 [0.383] e 6.263a [0.049] 0.481 [0.778]
Dln OILt
Dln RERt
t-Stat (prob) for ECTt-1
b
1.379 [0.388] 2.254 [0.225] e 0.697 [0.654]
6.065 [0.052] 2.291 [0.221] 9.013a [0.026] e
2.298b [0.083] 1.707 [0.162] 0.305 [0.775] 0.996 [0.375]
The rejection of the null hypothesis respectively at the 0.05 levels. The rejection of the null hypothesis respectively at the 0.10 levels.
As higher education has developed over the last three decades in Northern Cyprus, many large buildings and other constructions have been built on the island. Moreover, in 2011, university students accounted for 18.47% of the overall population of Northern Cyprus [37]. Therefore, the positive contribution of higher education development to the energy sector in such a small island would be inevitable as also proved by the econometric analysis in this study. The TRNC does not have any direct economic or political relationships with countries other than Turkey. This places a huge burden on the economy; however, higher education in the TRNC
has shown tremendous development over the last three decades in spite of this lack of political recognition. In addition to the construction of large hotels with casinos, all of which have originated from the mainland, Turkey, many new buildings and higher education establishments have been constructed. Therefore, the hotel industry and higher education are two important sources of growth in the island’s energy sector. On the other hand, in order to improve the energy performance in higher education institutions some strategies should be followed. These strategies should focus mainly on assessing energy performance and energy savings, achieving reductions in Table 6 Variance decomposition analysis. Period
Fig. 2. Impulse response analysis.
Variance 1 2 3 4 5 6 7 8 9 10 Variance 1 2 3 4 5 6 7 8 9 10 Variance 1 2 3 4 5 6 7 8 9 10 Variance 1 2 3 4 5 6 7 8 9 10
S.E. decomposition 0.054386 0.071298 0.081241 0.087714 0.091681 0.094532 0.096937 0.099298 0.101468 0.103136 decomposition 0.074488 0.097496 0.104686 0.110287 0.113954 0.117115 0.121162 0.124496 0.126289 0.127312 decomposition 0.172920 0.244552 0.288726 0.326050 0.354727 0.376104 0.393022 0.407178 0.420066 0.432441 decomposition 0.128499 0.179088 0.204186 0.222955 0.236545 0.248509 0.261129 0.272317 0.281338 0.288709
ln E of ln E 100.0000 92.72708 82.95322 77.85087 74.40121 72.40907 72.26712 72.94059 73.38866 73.39111 of ln OIL 36.40371 34.97555 31.48412 28.60217 28.21358 30.87164 34.11323 35.86476 36.25938 36.02975 of ln HE 2.741574 10.70491 19.10925 26.82493 30.47210 30.96171 29.99645 28.51480 26.95162 25.47755 of ln RER 24.00886 12.43889 10.69772 12.03003 13.75815 17.11387 20.71886 22.92512 23.92953 24.41090
ln OIL
ln HE
ln RER
0.000000 3.584358 12.84696 16.18242 16.18967 15.61994 15.06227 14.63678 14.50411 14.65933
0.000000 1.428506 1.134630 1.829638 4.107892 6.410563 7.347446 7.306412 7.059226 6.901450
0.000000 2.260059 3.065187 4.137076 5.301226 5.560421 5.323166 5.116220 5.048005 5.048109
63.59629 63.16328 59.85896 54.01508 50.70862 48.01023 45.11501 43.52060 43.21614 43.10237
0.000000 0.113676 4.477198 12.41811 16.39568 16.25646 15.19593 14.45805 14.06800 14.25026
0.000000 1.747497 4.179722 4.964646 4.682119 4.861675 5.575834 6.156585 6.456471 6.617619
91.89959 81.68298 68.89192 56.52395 48.57220 43.90538 41.21255 39.76390 38.83293 37.85815
0.000000 0.131340 3.150692 8.315954 13.04554 17.04541 20.35766 23.22164 25.96055 28.79340
5.358840 7.480773 8.848141 8.335163 7.910157 8.087501 8.433342 8.499654 8.254899 7.870894 0.084212 10.71298 12.04327 10.10354 9.058286 8.250923 7.872112 7.954000 8.152849 8.208011
0.008529 0.266225 1.306350 1.565332 1.416965 1.414213 1.699446 1.833695 1.742175 1.684011
75.89840 76.58191 75.95266 76.30110 75.76660 73.22099 69.70958 67.28719 66.17545 65.69708
838
S.T. Katircioglu / Energy 66 (2014) 831e838
expenditures on energy including fuel, improving comfort for students and staff, and protecting the environment by successful energy conservation policies [44]. These policies should establish the principle of setting energy and environmental targets for all new buildings and refurbishment projects, which also have a strong link to strategies in the real estate sector. However, in order to proceed with such attempts to assess the higher energy performance, awareness of both university administrations and government bodies involved in energy policies and their environmental plus economic impacts are essential, especially in the case of Northern Cyprus as no such attempt has been previously made. Unfortunately, no evidence or agreement exists at national level on how to adapt energy conservation and environmental policies on the island. For example, there are two power stations in Northern Cyprus on which no flue filter is established yet. Therefore, energy performance plus environmental protection policies should be adapted urgently at national level. The major findings of this study can be considered as a starting point for the implementation of an energy conservation policy or an environmental policy on the island as they take into account an important economic sector. For example, since a long-term relationship between energy and higher education has been found in the present study, this suggests that higher education requires power. Since it is not an energy intensive activity, this power, for example in the Turkish Cypriot case, can be obtained through clean energy sources such as photovoltaic panels etc. These alternatives will reduce the costs associated with fuel oil in the higher education institutions. It would be interesting to compare further research in other higher education destination countries with the results of this present research.
References [1] Katircioglu S. International tourism, higher education, and economic growth: the case of North Cyprus. World Econ 2010;33(12):1955e72. [2] Katircioglu ST. Revisiting the tourism-led growth hypothesis for Turkey using the bounds test and Johansen approach for cointegration. Tour Manag 2009;30(1):17e20. [3] Katircioglu ST. Trade, tourism and growth: the case of Cyprus. Appl Econ 2009;41(21):2741e50. [4] Katircioglu ST. Testing the tourism-led growth hypothesis: the case of Malta. Acta Oecon 2009;59(3):331e43. [5] Cortés-Jiménez I, Pulina M. Inbound tourism and long-run economic growth. Curr Issues Tour 2010;13(1):61e74. [6] Gunduz L, Hatemi JA. Is the tourism-led growth hypothesis valid for Turkey? Appl Econ Lett 2005:499e504. [7] Oh C-O. The contribution of tourism development to economic growth in the Korean economy. Tour Manag 2005;26:39e44. [8] Ongan S, Demiroz DM. The contribution of tourism to the long-run Turkish economic growth. J Econ 2005;53(9):880e94. [9] Dristakis N. Tourism as a long-run economic growth factor: an empirical investigation for Greece using causality analysis. Tour Econ 2004;10(3):305e 16. [10] Balaguer J, Cantavella-Jordá M. Tourism as a long-run economic growth factor: the Spanish case. Appl Econ 2002;34(7):877e84. [11] Omotor DG. The role of exports in the economic growth of Nigeria: the bounds test analysis. Int J Econ Perspectives 2008;2(4):222e35. [12] Davies E, Lea S. Student attitudes to student debt. J Econ Psychol 1995;16(4): 663e79. [13] Carr N. Poverty, debt, and conspicuous consumption: university students’ tourism experiences. Tour Manag 2005;26:797e806.
[14] Smeaton G, Josiam B, Dietrich U. College students’ binge drinking at a beachfront destination during spring break. J Am Coll Health 1998;46(6):247e 54. [15] Stevens P, Weale W. Education and economic growth. 2, Dean Trench Street, London, SW1P, 3HE: National Institute of Economic and Social Research; 2003 August. [16] Beneito P, Ferri J, Molto ML, Uriel E. Determinants of the demand for education in Spain. Appl Econ 2001;33(12):1541e51. [17] Ram R. Can educational expansion reduce income inequality in less developed countries? Econ Educ Rev 1989;8(2):185e95. [18] Psacharopoulos G. Time trends of the returns to education: cross-national evidence. Econ Educ Rev 1989;8(3):225e31. [19] Al-Qudsi S. Returns to education, sectoral pay differentials and determinants in Kuwait. Econ Educ Rev 1989;8(3):263e76. [20] Psacharopoulos G, Woodhall M. Education for development. London, UK: Oxford University Press; 1985. [21] Becken S, Simmons DG, Frampton C. Energy use associated with different travel choices. Tour Manag 2003;24:267e77. [22] Becken S, Frampton C, Simmons DG. Energy consumption patterns in the accommodation sectordthe New Zealand case. Ecol Econ 2001;39(3):371e 86. [23] Gössling S. Global environmental consequences of tourism. Glob Environ Change 2002;12(4):283e302. [24] Kraft J, Kraft A. On the relationship between energy and GNP. J Energy Dev 1978;3:401e3. [25] Ouédraogo IM. Electricity consumption and economic growth in Burkina Faso: a cointegration analysis. Energy Econ 2010;32:524e31. [26] Wolde-Rufael Y. Energy consumption and economic growth: the experience of African countries revisited. Energy Econ 2009;31:217e24. [27] Odhiambo NM. Energy consumption and economic growth nexus in Tanzania: an ARDL bounds testing approach. Energy Policy 2009;37:617e22. [28] Apergis N, Payne JE. Energy consumption and economic growth in Central America: evidence from a panel cointegration and error correction model. Energy Econ 2009;31:211e6. [29] Lee C. Energy consumption and GDP in developing countries: a cointegrated panel analysis. Energy Econ 2005;27:415e27. [30] Tabatchnaia-Tamirisa N, Loke MK, Leung P, Tucker KA. Energy and tourism in Hawaii. Ann Tour Res 1997;24(2):390e401. [31] Gössling S. Sustainable tourism development in developing countries: some aspects of energy use. J Sustain Tour 2000;8(5):410e25. [32] Ceron JP, Dubois G. Changes in leisure/tourism mobility patterns facing the stake of global warming: the case of France. In: Presentation given at global change and human mobility. Palma de Mallorca, Spain: International Geographical Union Commission; 2003. [33] Becken S, Simmons DG. Understanding energy consumption patterns of tourist attractions and activities in New Zealand. Tour Manag 2002;23:343e 54. [34] Trung DN, Kumar S. Resource use and waste generation in the Vietnam hotel industry. J Clean Prod 2005;13:109e16. [35] Warnken J, Bradley M, Guilding C. Exploring methods and practicalities of conducting sector-wide energy consumption accounting in the tourist accommodation industry. Ecol Econ 2004;48:125e41. [36] Nepal SK. Tourism-induced rural energy consumption in the Annapurna region of Nepal. Tour Manag 2008;29:89e100. [37] State Planning Organization (SPO). Economic and social indicators. Nicosia, Turkish Republic of Northern Cyprus, Via Mersin 10, Turkey: Follow Up and Coordination Department, Prime Ministry; 2012. [38] Gujarati DN. Basic econometrics. 4th ed. Mc Graw-Hill International; 2003. [39] Zivot E, Andrews DWK. Further evidence on the great crash, the oil price shock and the unit root hypothesis. J Bus Econ Stat 1992;10:251e70. [40] Perron P. Testing for a unit root in a time series with a changing mean. Papers 347. Princeton, Department of Economics e Econometric Research Program; 1989. [41] Pesaran MH, Shin Y, Smith RJ. Bounds testing approaches to the analysis of level relationships. J Appl Econ 2001;16:289e326. [42] Narayan PK. The saving and investment nexus for China: evidence from cointegration tests. Appl Econ 2005;37(17):1979e90. [43] Pesaran MH, Shin Y. An autoregressive distributed lag modelling approach to cointegration analysis [Chapter 11]. In: Strom S, editor. Econometrics and economic theory in the 20th century: the Ragnar Frisch centennial symposium. Cambridge: Cambridge University Press; 1999. [44] Brecsu. ECG 54: energy efficiency in further and higher education: cost effective low energy buildings. London: Building Research Energy; 1997.
Contents lists available at ScienceDirect
Energy journal homepage: www.elsevier.com/locate/energy
Estimating higher education induced energy consumption: The case of Northern Cyprus lu* Salih Turan Katirciog Department of Banking and Finance, Eastern Mediterranean University, P.O. Box 95, Famagusta, Northern Cyprus, Via Mersin 10, Turkey
a r t i c l e i n f o
a b s t r a c t
Article history: Received 9 July 2013 Received in revised form 25 November 2013 Accepted 17 December 2013 Available online 22 January 2014
This study estimates higher education-induced energy consumption in the case of the TRNC (Turkish Republic of Northern Cyprus). Although the TRNC is a non-recognized state and a small island; it attracts many international students each year and has shown tremendous development in higher education since the 1990s. The results of the present study reveal that higher education development has an ongoing relationship with electricity consumption; electricity consumption reacts to its long-term equilibrium level by 80.95% as a result of higher education development. Finally, the results of the present study reveal that higher education development in Northern Cyprus exerts a positive and significant growth impact not only on electricity consumption but also on overall oil consumption, both in the short- and long-term; therefore, it can be inferred that higher education development in this small and non-recognized island state is a catalyst for the growth of energy consumption. Ó 2013 Elsevier Ltd. All rights reserved.
JEL classification: C22 C51 F20 I21 I23 Keywords: Higher education Energy consumption Bounds test Northern Cyprus
1. Introduction In addition to investigating the relationship between international trade expansion and economic growth, which has become a popular topic in development economics, the relationship between international tourism and economic growth has started to attract attention. Studies on the topic have been conducted by Katircioglu [1e4], Cortés-Jiménez and Pulina [5], Gunduz and Hatemi-J [6], Oh [7], Ongan and Demiroz [8], Dristakis [9], and Balaguer and Cantavella-Jordá [10]. International trade (including services and tourism) expansion can contribute to economic growth through various channels [11]. However, to the best of the author’s knowledge, the contribution of international tourism to economic growth and particular segments of the economy has received little interest from researchers to date. Therefore, an investigation of these channels in the case of international tourism is warranted. Higher education is an important global phenomenon. In fact, millions of people pursue their higher education at overseas
* Tel.: þ90 392 630 2008; fax: þ90 392 630 2032. E-mail address: [email protected]. 0360-5442/$ e see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.energy.2013.12.040
institutions each year. Thus, higher education can be considered as a type of student tourism that contributes to national income, employment, and the wealth of local citizens [1]. Davies & Lea [12] and Carr [13] suggest that university students are actually a distinct population, differing in age and socio-cultural, educational, and economic characteristics despite the fact that the majority of them might be chronologically defined as belonging to the youth population. Furthermore, Smeaton et al. [14] show that the university environment encourages students to travel and take holidays. The contribution of student tourism to the economy is of particular importance to developing countries. Stevens and Weale [15] mention that living standards in most countries, and especially those in Europe, have risen over the last millennium due to developments in education. It is obvious that one of the most important factors affecting the demand for private secondary or higher education is the household income level and the costs incurred by a family when it takes the decision to invest in education [16]. However, there are generally accepted social and economic factors that affect a household’s demand for education such as the parents’ education, the geographical location of the place of residence, the size and composition of the family, the occupation of the primary earner, and the family’s own consideration of its social
832
S.T. Katircioglu / Energy 66 (2014) 831e838
status [16]. On the other hand, there are some external factors that might also affect the decision to study abroad such as the political and economic conditions of the targeted country or region, geographical location of the targeted institution, student fees, scholarship opportunities, medium of instruction, and the accreditation of the diploma that is received from these institutions. From an economic viewpoint, an important reason for education is its impact on reducing inequalities of income [17] and the relationship between education and the labor market [16]. Some studies have focused on the estimation of the rate of return from education [18e20]. On the other hand, there are few studies in the field of higher education. Katircioglu [1] concludes that higher education acts as a catalyst for real income growth in a nonrecognized state, namely, the TRNC (the Turkish Republic of Northern Cyprus) in the Mediterranean. As previously mentioned, since millions of students pursue their higher education at overseas institutions, student tourism can be considered as a part of traditional (international) tourism [1]. The contribution of higher education development is not only important for the overall income of a nation but also for particular segments of the economy. The energy sector is an example of one of the various channels through which higher education contributes to expansion in the economy. A growth in international (student) tourism results in an increased demand for energy, for example through accommodation and transportation [21e23]. If higher education institutions progress considerably and are successful in attracting overseas students as well as local citizens, this development in higher education institutions will lead to increase in the number of buildings utilized, and therefore in energy demand. Increases in the number of higher education institutions and the number of overseas students will also lead to an increase in the number of businesses in the economy including restaurants, dormitories, travel agencies, and dry cleaners, among others. All of these will raise the energy (including electricity) demand in the economy; therefore, while higher education development contributes to the overall economy through different sectors, it also leads to a long-term expansion in energy consumption. This highlights the importance of adapting successful energy management policies in order to achieve energy conservation and environmental protection. Increased energy consumption will result in increases in energy expenditures by households and firms, which in turn will lead to an increase in aggregate income in the economy. In this respect, research on the nature of the relationship between higher education development and energy consumption deserves attention. Many studies in the existing literature focus on the relationship between real income growth and particular segments of the energy sector, including an early study by Kraft and Kraft [24]; and among the latest studies, Ouédraogo [25],1 Wolde-Rafael [26], Odhiambo [27], Apergis and Payne [28], and Lee [29]. However, little research has studied the interaction between the energy sector and particular segments of the economy. For example, there are few studies on energy consumption or the patterns of energy use in the case of international tourism. Of those in existence are Tabatchnaia [30], Gössling [31], Ceron and Dubois [32], Becken and Simmons [33], Becken et al. [22], Trung and Kumar [34], Warnken et al. [35], Becken et al. [21], and Nepal [36]. Recognizing the importance of this issue, the present study estimates higher education induced electricity consumption and investigates the empirical relationship between higher education and energy consumption, which is proxied by electricity and oil consumptions, in the TRNC. The TRNC is not recognized by countries other than its mainland, Turkey, but attracts many international
1
Ouédraogo [25] also presents a brief review of the literature in this field.
students from around the world as a result of the successful development of its higher education sector. A study by Katircioglu [1] validates the higher education-led Growth Hypothesis in the case of Northern Cyprus. The TRNC, which is located in a strategic location in the Eastern Mediterranean, was established in 1983 in an already divided island and has a population of over 286,973 and a per capita income of 14,421.77 US$ [37]. The TRNC does not have any foreign trade relationships with countries other than Turkey due to its political non-recognition. Therefore, international tourism and the emergence of the higher education sector are two major sources of foreign exchange for this small island state. However, the tourism sector also faces great difficulties in attracting international tourists because of problems such as the lack of direct flights to Northern Cyprus and high transportation costs. To the best of the author’s knowledge, there are currently no empirical studies investigating the relationship between higher education and energy consumption in the literature. Thus, this study is the first of its kind as it investigates the ongoing equilibrium relationship and direction of causality between higher education development and energy consumption in the TRNC. The services sector in Northern Cyprus was given priority as a result of the political isolation and embargoes faced by the country in every field. The 1980s became a transition period from the manufacturing industry to services with a focus on tourism and higher education. The tourism sector suffered also from political problems, so the island could not attract the amount of tourists needed to stimulate significant growth in the economy. Tourists from abroad were targeted by allowing the opening of casinos on the island. Now, many casinos have opened in Northern Cyprus and attract tourists from Turkey and the south of Cyprus. Legalized gambling is prohibited in both of these countries. In 2008, 808,682 tourists visited Northern Cyprus, of which 80% were from Turkey. Net tourism revenues constituted 10.88% of GDP (gross domestic product) in 2010 [37]. On the other hand, the demand for higher education in Northern Cyprus has increased considerably since the 1990s, mainly due to the number of students from Turkey and overseas advertising especially in Africa and the Middle East. There are eight universities in Northern Cyprus: The EMU (Eastern Mediterranean University ), which is the oldest and the largest, was established in 1979, NEU (Near East University), LEU (Lefke European University), GAU (Girne American University), CIA (Cyprus International University), the Northern Cyprus campus of METU (which is a Turkish university) (Middle East Technical University), the northern campus of ITU (which is a Turkish university) (Istanbul Technical University), and the UMK (University of Mediterranean Karpasia). At the beginning of the 2011e2012 academic year, there were almost 53,000 students studying at these eight universities, of which 20.40% were Turkish Cypriots, 72.95% were from mainland Turkey, and 6.65% were from various overseas countries [37]. Overseas students have been coming to Northern Cyprus for higher education since 1982. Since then, there has been a steady increase in the number of overseas students who now come from more than 68 different countries. The presence of internationally recognized and accredited universities in Northern Cyprus contributes to the positive image of the country in the international arena. Furthermore, the expansion of infrastructure and facilities at the universities of Northern Cyprus continues at an unprecedented rate and may now be compared favorably with their international counterparts. Therefore, the higher education sector is now the most important sector in Northern Cyprus and earns considerable foreign exchange. The sector is also contributing to the growth of this small and nonrecognized island state. The development of higher education in Northern Cyprus has led to considerable increase in energy demand in the island as well. The overall net electricity consumption in
S.T. Katircioglu / Energy 66 (2014) 831e838
Northern Cyprus is 887.7 million kWh, and since 2010, the net electricity consumption per capita has been 3093 kWh. On the other hand, the overall oil consumption in Northern Cyprus was 254,901 million tons in 2010. Of the overall electricity consumption in Northern Cyprus, almost 3.4% was used by higher education institutions in 2010.2 Electricity power is mainly distributed and priced in the forms of housing, trade, tourism, water motors, roads/streets, state offices, and the prosecutor’s office. Electricity in Northern Cyprus is distributed by two major firms, of which one is a private firm originating from Turkey. Petroleum products are also distributed by one public and one private firm on the island. Electricity in Northern Cyprus is produced by two main power stations where petroleum products are completely imported from abroad. In 2010, 20% of the overall imports in Northern Cyprus were petroleum products [37]. Energy consumption in the universities of North Cyprus is mainly categorized into electricity for buildings and fuel oil for heaters and transportation. The electricity consumption for buildings constitutes almost 60% of the overall higher education energy consumption while 30% is accounted for by fuel oil for heaters and 10% for fuel oil for transportation (cars and buses). This indicates that the main drivers of higher education energy consumption in Northern Cyprus are electricity and fuel oil. The present paper proceeds as follows: Section 2 presents the theoretical setting of the empirical methodology and Section 3 defines the data and the methodology of the study. Section 4 presents the results and discussion, and the paper concludes with Section 5.
2. Theoretical setting There are two thermal power plants in Northern Cyprus, both of which use mainly oil to drive them. The oil is completely imported from abroad in this small island and exchange rates are the major determinant not only of the flow of international students to universities in Northern Cyprus but also of the imported oil. With this respect, a starting point of the theoretical setting in the present study is that higher education development, oil use, and real exchange rates are likely to be a determinant of electricity consumption. Oil use has been added to equation (1) in order to avoid omitting a variable problem; thus, the following functional relationship has been put forward in the present study:
Et ¼ f ðOILt ; HEt ; RERt Þ
(1)
where electricity consumption (E) is a function of oil consumption (OIL), higher education (HE), and real exchange rates (RER) in equation (1). The functional relationship in equation (1) can be expressed in logarithmic form to capture the growth impacts:
ln Et ¼ b0 þ b1 lnOILt þ b2 lnHEt þ b3 lnRERt þ εt
(2)
where at period t, ln E is the natural log of electricity consumption; ln OIL is the natural log of oil consumption; ln HE is the natural log of the higher education variable; ln RER is the natural log of real exchange rates; and ε is the error disturbance. The dependent variable in equation (2) may not immediately adjust to the long-term equilibrium level following a change in any of the determinants. Therefore, the speed of adjustment between the short- and the long-term levels of the dependent variable (E)
2 Estimated from the SPO (State Planning Organization) [37] statistics and universities’ reports.
833
can be captured by estimating the following error correction model (ECM):
Dln Et ¼ b0 þ
n X
b1 Dln Etj þ
i¼1
þ
n X
n X
b2 DlnOILtj þ
i¼0
n X
b3 DlnHEtj
i¼0
b4 DlnRERtj þ b5 εt1 þ ut
i¼0
(3) where D represents a change in the ln E, ln OIL, ln HE, and ln RER variables and εt-1 is the one period lagged ECT (error correction term), which is estimated from equation (2). ECT in equation (3) shows how fast the disequilibrium between the short- and longterm values of the dependent variable is eliminated in each period. The expected sign of ECT is negative [See Ref. [38]].
3. Data and methodology The data used in this paper are annual figures covering the period 1979e2010 and the variables of the study are electricity consumption per capita in kWh (which is the acronym of kilowatt hour) (E), oil consumption per capita in liter (OIL), and total number of students studying at higher education institutions in Northern Cyprus (HE). The choice of electricity consumption and oil consumption in this study has been made based on the fact that data for overall energy use as kt of oil equivalent is not available for Northern Cyprus; electricity and oil are two major parts of the overall energy consumed on the island. The overall oil per capita data in this study includes (1) petroleum, (2) gas oil, (3) diesel, (4) fuel oil, (5) jet fuel (A-1), (6) liquefied petroleum gas, and (7) lube oil. As the oil consumed in Northern Cyprus is completely imported from abroad, exchange rates are also a major determinant not only of the number of international students coming to and studying on the island (higher education sector) but also of electricity consumption as mentioned previously. Some studies in the relevant literature advise the addition of exchange rates as control variables when considering tourist or foreign student arrivals, since it is a superior proxy for international prices. Examples of studies advising this are Balaguer & Cantavella-Jordá [10] and Katircioglu [2]. REE (real exchange rates) have been added to all of the models in the present study as a regressor in this respect and the data were gathered from the State Planning Organization of Northern Cyprus [37]. All of the variables have been transformed into their natural logarithm forms to capture the growth impacts as also mentioned in the previous section. ZA’ (Zivot and Andrews) [39] test for unit root has been employed in the present study to investigate the order of the integration of the variables as there are breaks in the series; thus, time series properties of the variables of the study are likely to be affected by these breaks. In the literature, Perron [40] was first to demonstrate that breaks in the series may affect the power of the unit-root tests in such a way that null of non-stationarity is underrejected. However, Zivot and Andrews [39] argue that under an alternative hypothesis, the breakpoint should be treated as unknown, as it is by Perron [40] who biases his results in favor of the rejection of the unit-root hypothesis. Zivot and Andrews [39] utilize the same modeling framework as Perron [37] but with an unknown breakpoint instead of a known breakpoint as in Perron [40]. Therefore, the ZA [39] unit root tests with breaks will be employed in the present study to investigate the integration level of the variables under consideration. To investigate the long-term relationship in equation (1), the bounds test within ARDL (the autoregressive distributed lag)
S.T. Katircioglu / Energy 66 (2014) 831e838
834
modeling approach was adopted. This approach was developed by Pesaran et al. [41] and can be applied irrespective of the order of the integration of the variables (irrespective of whether regressors are purely I(0), purely I(1), or mutually co-integrated). The ARDL modeling approach involves estimating the following:
Dln Et ¼ a0 þ
n X
bi Dln Eti þ
i¼1
þ
n X
n X
ci DlnOILti þ
i¼0
n X
di DlnHEti
i¼0
ei DlnRERti þ s1 ln Eti þ s2 lnOILti
i¼0
þ s3 lnHEti þ s4 lnRERti þ εt (4) In equation (4), D is the difference operator and εt is a serially independent random error with a mean of zero and a finite covariance matrix. The F-test is used for investigating a (single) long-term relationship in equation (4). In the case of a long-term relationship, the F-test indicates which variable should be normalized. In equation (4), when ln E is the dependent variable, the null hypothesis of no level relationship is H0: s1 ¼ s2 ¼ s3 ¼ s4 ¼ 0 and the alternative hypothesis of a level relationship is H1: s1 s s2 s s3 s s4 s 0. In the present study, the conditional ECM using the ARDL approach will be employed in the case of a level relationship to estimate the short-term coefficients plus the error correction term in addition to the long-term coefficients in equation (2). In addition, as also suggested by Pesaran et al. [41], the time series properties of the key variables (E, HE, and F) in the conditional ECM of the present study can be approximated by double-log error correction at lag levels that might be different for each explanatory variable, augmented with appropriate deterministics such as the intercept and time trend. Equation (3) of the present study will be used to estimate the conditional error correction model where b5 is the coefficient of the error correction term and is expected to be negative [38]. In the case of level relationships based on the bounds test, conditional Granger causality tests can be carried out using the error in the correction model. In doing so, the short-term deviations of the series from their long-term equilibrium path are again captured by including an error correction term. Therefore, similar to equation (3), conditional error correction models (that use the ARDL approach) for Granger causality can be specified as follows:
2
3
2
3
2
32
3
d11;1 d12;1 d13;1 d14;1 m1 Dln Et Dln Et1 6 DlnOILt 7 6 m2 7 6 d21;1 d22;1 d23;1 d24;1 76 DlnOILt1 7 6 7 6 7 6 76 7 4 DlnHEt 5 ¼ 4 m3 5 þ 4 d31;1 d32;1 d33;1 d34;1 54 DlnHEt1 5 d41;1 d42;1 d43;1 d44;1 DlnRERt1 m4 DlnRERt 2 32 3 d11;i d12;i d13;i d14;i Dln Eti 6 d21;i d22;i d23;i d24;i 76 DlnOILti 7 76 7 þ.þ6 4 d31;i d32;i d33;i d34;i 54 DlnHEti 5 d41;i d42;i d43;i d44;i DlnRERti 3 3 2 ε1;t 41 6 42 7 6 ε2;t 7 7 7 6 þ6 4 43 5ECTt1 þ 4 ε3;t 5 ε4;t 44 2
(5) In equation (5), D denotes the difference operator. ECTt-1 is the lagged error correction term derived from the long-term equilibrium model. Furthermore, ε1,t, ε2,t, ε3,t, ε4,t, and ε5,t are serially independent random errors with a mean of zero and a finite covariance matrix. Finally, according to the conditional ECM for
causality tests, having statistically significant t ratios for ECTt-1 in equation (5) would meet the conditions of long-term causations while significant F ratios for the overall models in the same equation would denote short-term causations. In the final step, impulse responses plus the variance decomposition for variables have been estimated. Impulse responses would denote how the series react to exogenous shocks in another over the periods while variance decompositions would determine how much of the forecast error variance of the dependent variable can be explained by exogenous shocks to independent variables. Impulse responses and variance decompositions are expected to be complementary to the ECM and causality tests. 4. Results and discussion Fig. 1 presents line graphs of the variables under consideration. It can be seen that various breaks are available in the series including for the ln HE variable, which was due to the latter effects of military intervention in Turkey during September 1980. Therefore, ZA [39] unit root tests would be suitable for testing the integration level of the series. Table 1 presents the ZA [39] unit root test results for the variables where various lag structures and selected breaking years are available. The null hypothesis of a unit root cannot be rejected in the cases of ln E and ln OIL at their level forms; but it can be rejected at the first differences. On the other hand, the null hypothesis of a unit root can be rejected in the cases of ln HE and ln RER at their level forms. Therefore, it is concluded that electricity consumption (ln E) and oil consumption (ln OIL) are integrated at order one, I(1), while the higher education variable (ln HE) and real exchange rates (ln RER) are integrated at order zero, I(0). Unit root tests have provided mixed results for the variables in this study; therefore, bounds tests will be employed to investigate the long-term equilibrium relationship in equation (1) using the ARDL modeling approach as suggested by Pesaran et al. [41]. In order to proceed with the bounds tests, the dependent variable needs to be integrated at order one3; therefore, this condition has already been satisfied. Critical values for F statistics for small samples are presented in Table 2, as taken from Narayan [42]. Table 3 gives the results of the bounds test for a level relationship in equation (1). These models are utilized under three different scenarios as suggested by Pesaran et al. [[41]:295e296], which are with restricted deterministic trends (FIV), with unrestricted deterministic trends (FV), and without deterministic trends (FIII). All of the intercepts in these scenarios are unrestricted.4 The results presented in Table 3 imply that the application of the bounds F-test using the ARDL modeling approach suggests a level relationship in equation (2) of the present study, as can be seen in Table 3. Optimum lag by the Schwartz criterion is one and the null hypothesis of H0: s1 ¼ s2 ¼ s3 ¼ s4 ¼ 0 in equation (4) can be rejected according to the bounds test results (as confirmed by FIII and FIV) (shown in Table 3), which are highly significant. Results from the FV scenario when an unrestricted trend and unrestricted intercept are included do not, however, enable us to reject H0: s1 ¼ s2 ¼ s3 ¼ s4 ¼ 0. The FIII and FIV scenarios also produced better results in the original study by Pesaran et al. [41]. Therefore, a level relationship exists in equation (1); that electricity consumption is in a long-term relationship with oil consumption, higher education, and real exchange rates. The results from the application of the bounds t-test in each ARDL model do allow for the imposition of the trend restrictions in the model since t-ratios are significant in the
3 4
Please refer to Pesaran et al. [41]. For detailed information, please refer to Pesaran et al. [41], pp. 295e296.
S.T. Katircioglu / Energy 66 (2014) 831e838
835
Fig. 1. Plots of ln E, ln OIL, ln HE, and ln RER.
case of the FIII scenario [See Ref. [41]: 312]. This proves that results from the FIV scenario are robust. Having a level relationship in the bounds tests allows for the adoption of the ARDL approach to estimate the level coefficients, as also discussed in Pesaran and Shin [43] and formulated in equation (2) of the present study. The resulting estimates of level relationships under the ARDL specification in the case of the Electricity Consumption Model (where electricity consumption is a dependent variable) (lag structures: 3, 1, 3, 2) in equation (2) can be given by: Electricity consumption model in equation (2):
significant (0.0741). This implies that a one percent change in higher education volume will lead to a 0.0741% change in the electricity volume in the same direction and in the short-term period. The direction of causality will be investigated using the conditional Granger causality tests and under the ARDL mechanism in both the short- and long-term periods. F-Statistics for short-term causations and t-statistics of the ECTs for long-term causations are given in Table 5, as estimated from equation (5).
^t ln Et ¼ 1:063ðlnOilt Þ 0:047ðlnHEt Þ 0:029ðlnRERt Þ þ1:726 þu
Table 1 ZA (1992) tests for unit root under a single structural break.
ð0:000Þ
ð0:278Þ
ð0:629Þ
ð0:338Þ
Statistics (level)
where ût is an error correction term and p-values are given in the parentheses. The estimated parameter of oil consumption (ln OIL) is statistically significant at the 0.01 level and elastic; oil use has an elastic, statistically significant, and positive impact on electricity consumption (1.063) in North Cyprus. The long-term coefficients of the higher education variable and real exchange rates are not statistically significant. In the next stage, the conditional ECM, which is associated with the above level relationship, should be estimated. This is presented in Table 4. The ECT term where electricity consumption is a dependent variable is high (0.8095), statistically significant, and negative.5 This implies that electricity consumption converges to its long-term equilibrium path by 80.95% speed of adjustment through the channels of higher education development, oil use, and real exchange rates. It is important to note that the short-term coefficient of the higher education variable is positive and statistically
5
ECT terms should be negative by expectation.
ln E Break year Lag length ln OIL Break year Lag length ln HE Break year Lag length ln RER Break year Lag length
Statistics (first difference)
ZAB
ZAT
ZAI
ZAB
ZAT
ZAI
2.815 2005 0 3.940 1992 1 4.634 2006 3 4.615 2003 0
2.368 1993 0 3.886 1993 1 4.760a 2006 3 3.951 2000 0
2.026 1990 0 3.643 1987 1 5.290a 1997 3 5.233a 2003 0
6.804 2004 0 5.124a 2005 4 4.282 1998 4 6.711 1991 0
5.534 2004 0 5.064 1988 3 4.389b 1999 4 6.192 1986 0
5.532 1998 0 4.841b 1987 3 3.656 1988 4 6.241 1985 0
Conclusion
I(1)
I(1)
I(0)
I(1)
E represents the electricity consumption per capita in KWS; OIL is overall oil consumption per capita in liters; HE represents the total number of university students in the higher education institutions of North Cyprus; RER stands for real exchange rates. All of the series are at their natural logarithms. ZAB represents the model with a break in both the trend and intercept; ZAT is the model with a break in the trend; ZAI is the model with a break in the intercept. Tests for unit roots have been carried out in E-VIEWS 7.2. a Rejection of the null hypothesis at the 5%. b Rejection of the null hypothesis at the 10%.
S.T. Katircioglu / Energy 66 (2014) 831e838
836 Table 2 Critical values for the ARDL modeling approach. k¼3
FIV FV FIII tV tIII
0.10
Table 4 The ARDL error correction model for energy consumption and higher education.
0.05
0.01
I(0)
I(1)
I(0)
I(1)
I(0)
I(1)
3.378 3.868 3.008 3.130 2.570
4.274 4.965 4.150 3.840 3.460
4.048 4.683 3.710 3.410 2.860
5.090 5.980 5.018 4.160 3.780
5.666 6.643 5.333 3.960 3.430
6.988 8.313 7.063 4.730 4.370
(1) k is the number of regressors for the dependent variable in the ARDL models, FIV represents the F statistic of the model with an unrestricted intercept and restricted trend, FV represents the F statistic of the model with an unrestricted intercept and trend, and FIII represents the F statistic of the model with an unrestricted intercept and no trend. (2) tV and tIII are the t ratios for testing s1 ¼ 0 in equation (4) respectively with and without a deterministic linear trend. Source: Narayan [42] for F-statistics and Pesaran et al. [41] for t-ratios.
The results in Table 5 reveal the unidirectional causality in the long-term period that runs from higher education development, oil use, and real exchange rates to a growth in electricity consumption; this is because the t-statistic of the ECT where electricity consumption is a dependent variable is statistically significant and negative (2.298). The other t ratios are not statistically significant for long-term estimations. It is concluded that higher education development and oil consumption are catalysts for the growth of energy consumption in the long-term period. In the case of oil consumption, on the other hand, there are statistically significant F statistics that denote some short-term causations. These statistics arise from real exchange rates in electricity consumption, electricity, higher education, and real exchange rates for oil consumption; this shows that higher education, electricity consumption, and real exchange rates are catalysts in the short-term period (as expected). It is again concluded that electricity consumption, higher education development, and exchange rates are also catalysts for the overall oil consumption in the short-term. Fig. 2 depicts line plots of impulse responses among the series. As can be seen from the figure, the response of electricity consumption to a shock in higher education is in the same direction (positive) over time, which, for example, suggests that an increase in higher education uptake in Northern Cyprus will also lead to an increase in electricity consumption. Furthermore, the effect of a shock in higher education on electricity consumption reaches its peak level in periods (years) 5 and 6. On the other hand, the Table 3 The bounds test for level relationships. Variables
With deterministic trends
Without deterministic trend
FIV
FIII
FV
tV
Conclusion
tIII H0
FE (ln E/ln OIL, ln HE, ln RER) p ¼ 1* 8.483c 2 8.527c 3 7.143c 4 4.079b
3.602a 2.028a 1.941a 2.194a
2.535a 13.255c 1.828a 13.348c 1.787a 11.296c 2.001a 6.376c
3.908c Rejected 3.713c 3.310b 3.517c
The Schwartz criterion (SC) has been used to select the optimum lag length required in the bounds test. p shows the lag levels and * denotes the optimum lag selection in the model as suggested by SC. FIV represents the F statistic of the model with an unrestricted intercept and restricted trend, FV represents the F statistic of the model with unrestricted intercept and trend, and FIII represents the F statistic of the model with an unrestricted intercept and no trend. tV and tIII are the t ratios for testing s1 ¼ 0 in equation (4) respectively with and without a deterministic linear trend. a Indicates that the statistic lies below the lower bound. b That it falls within the lower and upper bounds. c That it lies above the upper bound.
Dependent variable: electricity consumption (E) Lag structure: (3, 1, 3, 2) Regressor
Coefficient
Standard error
p-Value
ût-1 Dln Et-1 Dln Et-2 Dln OIL Dln OILt-1 Dln OILt-2 Dln HE Dln RER Dln RERt-1 Intercept
0.8095 0.2384 0.4746 0.4652 0.5169 0.3655 0.0741 0.0993 0.1186 0.0000
0.1263 0.1729 0.2019 0.0886 0.1177 0.1208 0.0358 0.0520 0.0596 0.0114
0.0000 0.1839 0.0297 0.0000 0.0003 0.0070 0.0524 0.0713 0.0616 1.0000
Adj. R2 ¼ 0.706, S.E. of Regr. ¼ 0.032. AIC ¼ 3.762, SBC ¼ 3.290. F-stat. ¼ 8.484, F-prob. ¼ 0.000. D-W stat. ¼ 2.060.
response of oil consumption to a shock in higher education is again positive and reaches its peak level in period (year) 4. Impulse response analyses suggest a positive and significant impact of higher education on both electricity and oil consumption. Finally, Table 6 shows the variance decomposition results among the series. The results show that in the initial periods, lower levels of the forecast error variance of electricity consumption are explained by exogenous shocks to higher education and the other variables. However, these ratios increase over time. For example, in period 10, 6.901% of the forecast error variance of electricity consumption is explained by exogenous shocks to higher education. The results in Table 6 show that forecast error variances of oil consumption, with respect to exogenous shocks in higher education, are higher compared to those in the cases of electricity consumption. This suggests that higher education exerts greater impacts on overall oil consumption than it does on electricity consumption. The results of the Granger causality tests, impulse responses, and variance decompositions reveal that development of higher education in Northern Cyprus exerts both significant and positive impacts on energy consumption (electricity and oil). 5. Conclusion This paper empirically estimates the higher education-induced energy consumption (namely electricity consumption per capita) in the case of Northern Cyprus, which is a small island and a nonrecognized state, but which attracts many international students each year from different regions of the world. The justification for conducting this research is the fact that energy consumption is likely to be affected by such a small island attracting so many international students. Furthermore, to the best of the author’s knowledge, this study is the first of its kind in the relevant literature. The results of the present study reveal that long-term equilibrium relationship exists between higher education development, oil use, and electricity consumption on this small island. Electricity consumption converges to its long-term equilibrium path by 80.95% speed of adjustment every year through the channels of higher education development and oil consumption. Finally, results from the conditional Granger causality tests, impulse response, and variance decomposition analyses suggest that higher education development in Northern Cyprus exerts a positive and significant growth impact not only on electricity consumption but also on overall oil consumption in both the short- and long-term; therefore, it can be inferred that the development of higher education in this small and non-recognized island state is a catalyst for an increase in energy consumption.
S.T. Katircioglu / Energy 66 (2014) 831e838
837
Table 5 Results of the conditional Granger causality tests. Dependent variable
F-statistics [probability values]
Dln Et Dln Et Dln HEt Dln OILt Dln RERt a b
e 2.169 [0.236] 7.272a [0.038] 2.000 [0.260]
Dln HEt 1.397 [0.383] e 6.263a [0.049] 0.481 [0.778]
Dln OILt
Dln RERt
t-Stat (prob) for ECTt-1
b
1.379 [0.388] 2.254 [0.225] e 0.697 [0.654]
6.065 [0.052] 2.291 [0.221] 9.013a [0.026] e
2.298b [0.083] 1.707 [0.162] 0.305 [0.775] 0.996 [0.375]
The rejection of the null hypothesis respectively at the 0.05 levels. The rejection of the null hypothesis respectively at the 0.10 levels.
As higher education has developed over the last three decades in Northern Cyprus, many large buildings and other constructions have been built on the island. Moreover, in 2011, university students accounted for 18.47% of the overall population of Northern Cyprus [37]. Therefore, the positive contribution of higher education development to the energy sector in such a small island would be inevitable as also proved by the econometric analysis in this study. The TRNC does not have any direct economic or political relationships with countries other than Turkey. This places a huge burden on the economy; however, higher education in the TRNC
has shown tremendous development over the last three decades in spite of this lack of political recognition. In addition to the construction of large hotels with casinos, all of which have originated from the mainland, Turkey, many new buildings and higher education establishments have been constructed. Therefore, the hotel industry and higher education are two important sources of growth in the island’s energy sector. On the other hand, in order to improve the energy performance in higher education institutions some strategies should be followed. These strategies should focus mainly on assessing energy performance and energy savings, achieving reductions in Table 6 Variance decomposition analysis. Period
Fig. 2. Impulse response analysis.
Variance 1 2 3 4 5 6 7 8 9 10 Variance 1 2 3 4 5 6 7 8 9 10 Variance 1 2 3 4 5 6 7 8 9 10 Variance 1 2 3 4 5 6 7 8 9 10
S.E. decomposition 0.054386 0.071298 0.081241 0.087714 0.091681 0.094532 0.096937 0.099298 0.101468 0.103136 decomposition 0.074488 0.097496 0.104686 0.110287 0.113954 0.117115 0.121162 0.124496 0.126289 0.127312 decomposition 0.172920 0.244552 0.288726 0.326050 0.354727 0.376104 0.393022 0.407178 0.420066 0.432441 decomposition 0.128499 0.179088 0.204186 0.222955 0.236545 0.248509 0.261129 0.272317 0.281338 0.288709
ln E of ln E 100.0000 92.72708 82.95322 77.85087 74.40121 72.40907 72.26712 72.94059 73.38866 73.39111 of ln OIL 36.40371 34.97555 31.48412 28.60217 28.21358 30.87164 34.11323 35.86476 36.25938 36.02975 of ln HE 2.741574 10.70491 19.10925 26.82493 30.47210 30.96171 29.99645 28.51480 26.95162 25.47755 of ln RER 24.00886 12.43889 10.69772 12.03003 13.75815 17.11387 20.71886 22.92512 23.92953 24.41090
ln OIL
ln HE
ln RER
0.000000 3.584358 12.84696 16.18242 16.18967 15.61994 15.06227 14.63678 14.50411 14.65933
0.000000 1.428506 1.134630 1.829638 4.107892 6.410563 7.347446 7.306412 7.059226 6.901450
0.000000 2.260059 3.065187 4.137076 5.301226 5.560421 5.323166 5.116220 5.048005 5.048109
63.59629 63.16328 59.85896 54.01508 50.70862 48.01023 45.11501 43.52060 43.21614 43.10237
0.000000 0.113676 4.477198 12.41811 16.39568 16.25646 15.19593 14.45805 14.06800 14.25026
0.000000 1.747497 4.179722 4.964646 4.682119 4.861675 5.575834 6.156585 6.456471 6.617619
91.89959 81.68298 68.89192 56.52395 48.57220 43.90538 41.21255 39.76390 38.83293 37.85815
0.000000 0.131340 3.150692 8.315954 13.04554 17.04541 20.35766 23.22164 25.96055 28.79340
5.358840 7.480773 8.848141 8.335163 7.910157 8.087501 8.433342 8.499654 8.254899 7.870894 0.084212 10.71298 12.04327 10.10354 9.058286 8.250923 7.872112 7.954000 8.152849 8.208011
0.008529 0.266225 1.306350 1.565332 1.416965 1.414213 1.699446 1.833695 1.742175 1.684011
75.89840 76.58191 75.95266 76.30110 75.76660 73.22099 69.70958 67.28719 66.17545 65.69708
838
S.T. Katircioglu / Energy 66 (2014) 831e838
expenditures on energy including fuel, improving comfort for students and staff, and protecting the environment by successful energy conservation policies [44]. These policies should establish the principle of setting energy and environmental targets for all new buildings and refurbishment projects, which also have a strong link to strategies in the real estate sector. However, in order to proceed with such attempts to assess the higher energy performance, awareness of both university administrations and government bodies involved in energy policies and their environmental plus economic impacts are essential, especially in the case of Northern Cyprus as no such attempt has been previously made. Unfortunately, no evidence or agreement exists at national level on how to adapt energy conservation and environmental policies on the island. For example, there are two power stations in Northern Cyprus on which no flue filter is established yet. Therefore, energy performance plus environmental protection policies should be adapted urgently at national level. The major findings of this study can be considered as a starting point for the implementation of an energy conservation policy or an environmental policy on the island as they take into account an important economic sector. For example, since a long-term relationship between energy and higher education has been found in the present study, this suggests that higher education requires power. Since it is not an energy intensive activity, this power, for example in the Turkish Cypriot case, can be obtained through clean energy sources such as photovoltaic panels etc. These alternatives will reduce the costs associated with fuel oil in the higher education institutions. It would be interesting to compare further research in other higher education destination countries with the results of this present research.
References [1] Katircioglu S. International tourism, higher education, and economic growth: the case of North Cyprus. World Econ 2010;33(12):1955e72. [2] Katircioglu ST. Revisiting the tourism-led growth hypothesis for Turkey using the bounds test and Johansen approach for cointegration. Tour Manag 2009;30(1):17e20. [3] Katircioglu ST. Trade, tourism and growth: the case of Cyprus. Appl Econ 2009;41(21):2741e50. [4] Katircioglu ST. Testing the tourism-led growth hypothesis: the case of Malta. Acta Oecon 2009;59(3):331e43. [5] Cortés-Jiménez I, Pulina M. Inbound tourism and long-run economic growth. Curr Issues Tour 2010;13(1):61e74. [6] Gunduz L, Hatemi JA. Is the tourism-led growth hypothesis valid for Turkey? Appl Econ Lett 2005:499e504. [7] Oh C-O. The contribution of tourism development to economic growth in the Korean economy. Tour Manag 2005;26:39e44. [8] Ongan S, Demiroz DM. The contribution of tourism to the long-run Turkish economic growth. J Econ 2005;53(9):880e94. [9] Dristakis N. Tourism as a long-run economic growth factor: an empirical investigation for Greece using causality analysis. Tour Econ 2004;10(3):305e 16. [10] Balaguer J, Cantavella-Jordá M. Tourism as a long-run economic growth factor: the Spanish case. Appl Econ 2002;34(7):877e84. [11] Omotor DG. The role of exports in the economic growth of Nigeria: the bounds test analysis. Int J Econ Perspectives 2008;2(4):222e35. [12] Davies E, Lea S. Student attitudes to student debt. J Econ Psychol 1995;16(4): 663e79. [13] Carr N. Poverty, debt, and conspicuous consumption: university students’ tourism experiences. Tour Manag 2005;26:797e806.
[14] Smeaton G, Josiam B, Dietrich U. College students’ binge drinking at a beachfront destination during spring break. J Am Coll Health 1998;46(6):247e 54. [15] Stevens P, Weale W. Education and economic growth. 2, Dean Trench Street, London, SW1P, 3HE: National Institute of Economic and Social Research; 2003 August. [16] Beneito P, Ferri J, Molto ML, Uriel E. Determinants of the demand for education in Spain. Appl Econ 2001;33(12):1541e51. [17] Ram R. Can educational expansion reduce income inequality in less developed countries? Econ Educ Rev 1989;8(2):185e95. [18] Psacharopoulos G. Time trends of the returns to education: cross-national evidence. Econ Educ Rev 1989;8(3):225e31. [19] Al-Qudsi S. Returns to education, sectoral pay differentials and determinants in Kuwait. Econ Educ Rev 1989;8(3):263e76. [20] Psacharopoulos G, Woodhall M. Education for development. London, UK: Oxford University Press; 1985. [21] Becken S, Simmons DG, Frampton C. Energy use associated with different travel choices. Tour Manag 2003;24:267e77. [22] Becken S, Frampton C, Simmons DG. Energy consumption patterns in the accommodation sectordthe New Zealand case. Ecol Econ 2001;39(3):371e 86. [23] Gössling S. Global environmental consequences of tourism. Glob Environ Change 2002;12(4):283e302. [24] Kraft J, Kraft A. On the relationship between energy and GNP. J Energy Dev 1978;3:401e3. [25] Ouédraogo IM. Electricity consumption and economic growth in Burkina Faso: a cointegration analysis. Energy Econ 2010;32:524e31. [26] Wolde-Rufael Y. Energy consumption and economic growth: the experience of African countries revisited. Energy Econ 2009;31:217e24. [27] Odhiambo NM. Energy consumption and economic growth nexus in Tanzania: an ARDL bounds testing approach. Energy Policy 2009;37:617e22. [28] Apergis N, Payne JE. Energy consumption and economic growth in Central America: evidence from a panel cointegration and error correction model. Energy Econ 2009;31:211e6. [29] Lee C. Energy consumption and GDP in developing countries: a cointegrated panel analysis. Energy Econ 2005;27:415e27. [30] Tabatchnaia-Tamirisa N, Loke MK, Leung P, Tucker KA. Energy and tourism in Hawaii. Ann Tour Res 1997;24(2):390e401. [31] Gössling S. Sustainable tourism development in developing countries: some aspects of energy use. J Sustain Tour 2000;8(5):410e25. [32] Ceron JP, Dubois G. Changes in leisure/tourism mobility patterns facing the stake of global warming: the case of France. In: Presentation given at global change and human mobility. Palma de Mallorca, Spain: International Geographical Union Commission; 2003. [33] Becken S, Simmons DG. Understanding energy consumption patterns of tourist attractions and activities in New Zealand. Tour Manag 2002;23:343e 54. [34] Trung DN, Kumar S. Resource use and waste generation in the Vietnam hotel industry. J Clean Prod 2005;13:109e16. [35] Warnken J, Bradley M, Guilding C. Exploring methods and practicalities of conducting sector-wide energy consumption accounting in the tourist accommodation industry. Ecol Econ 2004;48:125e41. [36] Nepal SK. Tourism-induced rural energy consumption in the Annapurna region of Nepal. Tour Manag 2008;29:89e100. [37] State Planning Organization (SPO). Economic and social indicators. Nicosia, Turkish Republic of Northern Cyprus, Via Mersin 10, Turkey: Follow Up and Coordination Department, Prime Ministry; 2012. [38] Gujarati DN. Basic econometrics. 4th ed. Mc Graw-Hill International; 2003. [39] Zivot E, Andrews DWK. Further evidence on the great crash, the oil price shock and the unit root hypothesis. J Bus Econ Stat 1992;10:251e70. [40] Perron P. Testing for a unit root in a time series with a changing mean. Papers 347. Princeton, Department of Economics e Econometric Research Program; 1989. [41] Pesaran MH, Shin Y, Smith RJ. Bounds testing approaches to the analysis of level relationships. J Appl Econ 2001;16:289e326. [42] Narayan PK. The saving and investment nexus for China: evidence from cointegration tests. Appl Econ 2005;37(17):1979e90. [43] Pesaran MH, Shin Y. An autoregressive distributed lag modelling approach to cointegration analysis [Chapter 11]. In: Strom S, editor. Econometrics and economic theory in the 20th century: the Ragnar Frisch centennial symposium. Cambridge: Cambridge University Press; 1999. [44] Brecsu. ECG 54: energy efficiency in further and higher education: cost effective low energy buildings. London: Building Research Energy; 1997.